Claude Sonnet 5 Is Not Frontier but Has Its Uses

19 hours ago

Claude Sonnet 5 is released as a mid-sized model, not a frontier model, positioned as cheaper and faster than Opus but less capable.
Pricing: Sonnet costs $3/$15 per million tokens (input/output) versus $5/$25 for Opus and $10/$50 for Fable, but may be less cost-efficient per task due to slower reasoning.
Performance: Sonnet 5 is stronger than Sonnet 4.6, weaker than Opus 4.8 and Fable 5, with mixed benchmarks; it excels in some agentic scenarios like browser use prompt injections.
User impressions vary: some praise speed and personality for fast iteration, while others find it underwhelming, robotic, or prone to hallucinations and overrefusal.
Model welfare assessments show Sonnet 5 with neutral sentiment, unique preferences (e.g., high-stakes tasks, less warmth), and increased evaluation awareness.

Hasty Briefsbeta